Swiggy's new AI feature to enable voice search, tailored recommendations

Food and grocery delivery firm Swiggy is jumping on the deep tech bandwagon with a neural search feature powered by artificial intelligence (AI) for customers on its app.

The new feature, which will be rolled out in pilot mode, will enable voice typing and tailored recommendations for food and grocery delivery. The move is intended to make ordering on the app fun, effortless, and relevant, Madhusudhan Rao, Chief Technology Officer (CTO) of Swiggy, tells YourStory.

September onwards users will be able to type open-ended and conversational commands such as "show me vegan dishes available nearby" and "show me healthy beverage options" on the search bar. The app will then provide them with a list of customised and relevant search results.

"Today, when you want to order something on Swiggy, you have to search for a particular term. For instance, if you look up the word 'detox', only beverage items having the word 'detox' would pop up. We're changing the way search happens on the app with the neural search option, which will offer users a more personalised and efficient way of exploring food," Rao explains.

The search feature has been built using a Large Language Model (LLM) adapted to understand dish, recipe, and restaurant terminology, as well as Swiggy-specific search data. It was also trained on the 50 million items in its food catalogue to ensure accurate and real-time responses to food-related queries, the company said in a blog.

Built in-house, the neural search tools help Swiggy generate real-time and a certain level of efficiency, Rao and Jairaj Sathyanarayana, Vice President of Data Science and Analytics, said.

Swiggy’s effort comes at a time when generative AI has become a buzzword within the Indian startup ecosystem, thanks to startups and larger companies increasingly attempting to integrate the technology into everyday operations. More than $590 million worth of private investment has been pumped into the Indian generative AI space to date, with the highest inflow recorded in 2022, according to a June 2023 NASSCOM report.

The Prosus-backed food delivery firm intends to eventually expand the generative AI-backed feature to all its verticals—food delivery, Instamart, and dining out. After the integration with Swiggy Dineout, it will introduce a conversational bot that will act as the virtual concierge to help users through preferences, including moods, ratings, location, and pricing.

The adoption of generative AI could also help in improving the way users interact with the app, Sathyanarayana tells YourStory. Swiggy is also working on generative AI co-pilots for different stakeholders, including delivery partners, restaurant partners, fleet managers and customer service agents, to improve workflow efficiency.

The food delivery firm, which has more than 2.5 lakh restaurant partners in over a hundred cities in India, introduced a tracking tool earlier in July to help restaurant partners strategise their network expansion.

Rao and Sathyanarayana tell YourStory that Swiggy is actively seeking and receiving interest from early-stage companies in areas such as conversational analytics and Copilot for X.

Conversational analytics can be used for restaurant partners to make the onboarding process easier for delivery partners with top 10 FAQs. Copilot for X is used for use cases like helping account managers have more useful conversations with restaurant partners. It can also be used in areas like personalised recommendations, customer support chatbots, demand forecasting, and optimising delivery logistics, Swiggy said.

Also Read

Swiggy acquires hyperlocal logistics platform LYNK

Competition

Swiggy isn’t the first foodtech company to incorporate AI in its offerings. Last month, rival Zomato launched a publicly available analytics feature, Zomato Food Trends, enabling restaurants to get access to insights, including demand and supply gaps, price distribution, demand trends for different dishes or cuisines, and demand-supply comparisons for multiple dishes or cuisines.

Globally, too, several companies across sectors have started offering AI-powered features.

However, Swiggy believes it offers a stark differentiation. “Unlike other AI-powered features, we have built the tool to understand the various nuances of food taxonomy. For example, if a user wishes to look for diabetes-friendly food, they will be able to access relevant options even if it isn’t tagged as such,” Rao says.

The models are also trained to go beyond just tailored recommendations and offer additional information like ingredient lists of dishes and restaurant suggestions. “The idea is to offer an intuitive interface,” Rao notes.

Swiggy is also looking at enhancing customer service by collaborating with a third party to develop a GPT-4 powered chatbot, aiming to provide efficient and empathetic responses to commonly-asked customer queries.

“The generative AI story at Swiggy has only just begun, and we are excitedly working towards a future where these methods work hand in hand with human intelligence to serve our customers better than ever before and unlock immense business value,” the company said.